14386
J. Phys. Chem. B 2006, 110, 14386-14390
Kinetics of Liquid Phase Semiconductor Photoassisted Reactions: Supporting Observations for a Pseudo-Steady-State Model Andrew Mills,*,† Jishun Wang,† and David F. Ollis‡ Department of Pure & Applied Chemistry, UniVersity of Strathclyde, Glasgow, G1 1XL, United Kingdom, and Chemical Engineering Department, North Carolina State UniVersity, Raleigh, North Carolina 27695-7905 ReceiVed: April 14, 2006; In Final Form: May 31, 2006
The kinetics of liquid phase semiconductor photocatalytic and photoassisted reactions are an area of some debate, reignited recently by an article by Ollis1 in which he proposed a simple pseudo-steady-state model to interpret the Langmuir-Hinshelwood type kinetics, commonly observed in such systems. In the current article, support for this model, over other models, is provided by a reinterpretation of the results of a study, reported initially in 1999,2 of the photoassisted mineralization of 4-chlorophenol, 4-CP, by titania films and dispersions as a function of incident light intensity, I. On the basis of this model, these results indicate that 4-CP is adsorbed more strongly on P25 TiO2 when it is in a dispersed, rather than a film form, due to a higher rate constant for adsorption, k1. In addition, the kinetics of 4-CP removal appear to depend on Iβ, where β ) 1 or 0.6 for when the TiO2 is in a film or a dispersed form, respectively. These findings are discussed both in terms of the pseudo-steady-state model and other popular kinetic models.
Introduction Semiconductor photoassisted reactions, especially ones in which the catalytic nature of the semiconductor has been fully established,3 i.e., photocatalytic reactions, continue to attract attention from both industry and academia. Several notable photocatalytic commercial products have emerged including the following: water purification systems, self-cleaning glass, and superhydrophilic tiles.4,5 The current high level of research activity is reflected by the fact that over 915 papers were published in 2005 alone with “photocatalyst” or “photocatalysis” in their titles, list of key words, or abstract. Much of this work has focused on the photocatalytic or photoassisted mineralization of organic pollutants, i.e. hν gEbg
organic + oxygen 9 8 H2O + CO2 + mineral acids SC (1) where Ebg is the band gap of the semiconductor, invariably TiO2.6 A great deal of the early work in this area centered on the use of dispersions of semiconductor particles, whereas nowadays there is increasing interest in semiconductor films produced by CVD or sol-gel methods.5,7 In much of this work the kinetics of reaction 1 gave a good fit to a LangmuirHinshelwood (LH) rate form, i.e.
kLHKapp ads [A]
-rA ) 1 + Kapp ads [A]
(2)
where -rA is the rate of removal of the reagent of interest, A (usually an organic pollutant), and kLH and Kapp ads are the apparent maximum rate of the photoreaction and the apparent Langmuir adsorption constant, respectively.8 It has been long *Address correspondence to this author. † University of Strathclyde. E-mail:
[email protected]. ‡ North Carolina State University. E-mail:
[email protected].
recognized that Kapp ads is not usually the dark Langmuir adsorp6,8,9 tion constant, Kads, and typically Kapp ads . Kads. An early rationale for all of the then known features of the LH rate form in semiconductor photoassisted reactions, based on four possible hydroxyl radical mechanisms as the slow step, was provided by Turchi and Ollis in 1990.8 The features accounted for included not only the observation that Kapp ads . Kads but also that kLH was a function of absorbed light intensity, Iβ, where β ) 1.0 and β ) 0.5 at low and high absorbed light intensities, respectively. A substantial reevaluation of the mechanism of semiconductor photoassisted reactions has had to take place in recent years, upon the observation that Kapp ads is, like kLH, light intensity dependent and possibly inversely related to kLH.6c,10 A simple mechanism that is able to account for all of the above has been proposed recently1 and is as follows: k1
A(liquid) 79 8 A(ads) k -1
kLH
A(ads) 98 products
(3) (4)
In this mechanism it is assumed that the key step 4 involves the reaction of the adsorbed reactant and a photogenerated surface hydroxyl radical. Since the latter will be a function of absorbed light intensity, which in turn will be proportional to the incident light intensity, I, it follows that kLH will also be a function of I. As suggested by Turchi and Ollis,8 it appears likely that kLH is related to I via the following expression:
kLH ) RIβ
(5)
where R is a proportionality constant and β is 1.0 or 0.5, as described earlier. In the most recent model,1 it is also assumed that a pseudo-steady-state hypothesis applies to the surface
10.1021/jp062317c CCC: $33.50 © 2006 American Chemical Society Published on Web 07/01/2006
Supporting Observations for a Pseudo-Steady-State Model
J. Phys. Chem. B, Vol. 110, No. 29, 2006 14387
coverage of the reactant, θA, so that:
θA )
TABLE 1: LH Parameters, kLH and Kapp ads , for the Photoassisted Destruction of 4-CP Mediated by TiO2 Film or a Dispersion
Kapp ads [A] 1 + Kapp ads [A]
(6)
where
Kapp ads
)
1/Kapp diss
)
k1 k-1 + RIβ
(7)
app and Kdiss is the apparent dissociation constant. Assuming that -rA ) kLHθA, it follows:
-rA )
kLH[A] Kapp diss + [A]
(8)
app where kLH is given by eq 5 and Kdiss by eq 7. This pseudoapp steady-state approach predicts that both kLH and Kdiss depend on the intensity of absorbed light raised to the same exponent.1 In support of the above mechanism, Ollis cited the results of three papers published in 2000, which revealed an intensity 10a dependence for Kapp ads when the reactant, A, was either phenol, 10b 10c acetophenone, or methyl viologen. However, two of the papers10b,c contained only three points or less and the third,10a an excellent study of the photomineralization of PhOH by TiO2, appeared to be a special case for the Ollis model, where RI . k-1 and the reaction approximates to adsorption followed by reaction. Encouragingly, further, possibly more striking, support for the pseudo-steady-state approach can be gleaned from the results of a study of light intensity effects on the kinetics of photomineralization of 4-chlorophenol, 4-CP, reported by Mills and Wang in 1999, in which the semiconductor sensitizer, Degussa P25 TiO2, was used as either a thin film or an aqueous dispersion.2 In the latter work the substantial body of results on the 4-CP/TiO2 system is not a special case and so can be used to test the Ollis model much more fully; this paper is a reanalyzis of those results,2 based on the pseudo-steady-state approach of Ollis.1
The Kinetics of the Photomineralization of 4-CP, Sensitized by TiO2 In the latter work,2 P25 titanium dioxide from Degussa was used as the photosensitizer either as a film, supported on the glass of the photoreactor, or as a dispersion (0.5 g dm-3). The aqueous reaction solutions under test (100 cm3; pH 2; (HClO4), containing different levels of 4-CP (0.1-1 mM)), were magnetically stirred (600 rpm) at least 15 min before, as well as throughout, any irradiation. The solutions were also continually purged with oxygen. Other work showed that the rate of reaction was independent of stirrer speed when any TiO2 film was used as the photosensitizer, eliminating the possibility that the observed kinetics were due to external mass transfer. The Thiele modulus for the system was evaluated as 0.043-0.028, implying little or no internal mass transfer effects for this system.11 HPLC was used to determine the initial rate of disappearance of 4-CP, i.e., -r, which was calculated over the first 20% of the reaction. Further details of the TiO2 film preparation and characterization are given in the original paper, along with a description of the glass Dreschel bottle photoreactor and surrounding irradiation system, comprising twelve, 8-W blacklight blue, UVA lamps with an aluminum reflector.2 The suitability of using 4-CP as a test reagent might be questioned initially given that, unlike PhOH, it can be photore-
TiO2 film
TiO2 dispersion
Irel
kLH (µM min-1)
Kapp ads (103 M-1)
kLH (µM min-1)
Kapp ads (103 M-1)
1 0.713 0.616 0.388 0.269 0.143 0.084
26.1((6.2) 19.2((2.2) 16.5((4.3) 10.6((2.1) 6.3((1.3) 2.9((0.3) 2.5((0.1)
1.70((0.41) 1.93((0.23) 1.89((0.51) 2.75((0.57) 3.32((0.74) 6.23((0.69) 5.09((0.25)
13.6((2.4) 11.2((1.5) 10.2((0.9) 8.1((0.5) 6.9((0.6) 3.9((0.3) 2.8((0.2)
4.52((0.87) 4.63((0.69) 4.27((0.40) 5.11((0.38) 5.00((0.50) 8.36((0.77) 9.08((0.74)
a Derived from studies of the initial rate of removal of 4-CP as a function of [4-CP] at different relative light intensities and use of eq 2 to extract values for kLH and Kapp abs at the different light intensities.
duced, as well as oxidized.12,13 However, the unambiguous intermediates associated with this process, such as PhOH and bicyclic compounds, are only observed at pH >7, and then at very low levels, since the initial hydroxyl phenyl radical is most likely formed at higher pH values.13 In this work, at pH 2, in 0.01 M HClO4 and O2-saturated solution, the major intermediates observed are 4-chlorocatechol (4-CC) and hydroquinone (HQ); both of which are produced promptly upon UV illumination, with the level of 4-CC usually 3 times that of HQ.14 The level of benzoquinone (BQ) generated is typically 17 times less than that of 4-CC, and although BQ can be produced via a reductive as well as oxidative mechanism, its generation appears to track the production of HQ, as expected if it were derived only via an oxidative pathway.14 Similar intermediates and relative quantities have been reported for the same photoassisted reaction by others,15,16 and this and the above results suggest that it is reasonable to assume that the photodegradation of 4-CP occurs predominantly via an oxidative pathway, as in the case of PhOH, for example.12 Another possible concern is wavelength selectivity, since a blacklight blue lamp, which emits 365 ( 17 nm light, is used here as the illumination source and others have noted some spectral selectivity over this wavelength range using the 4-CP/ TiO2 system.12 This does not appear to be an issue in this work since a separate experiment was performed in which the light from a BLB light, filtered with a 365 ( 5 nm interference filter, was used as the light source irradiating a 4-CP/TiO2 system. The results showed the intermediate product distribution was the same as that observed by using the full output of the BLB: a product distribution dominated by 4-CC and indicative of a photooxidative mechanism. Stafford et al. using either a 450-W medium-pressure Hg lamp or a BLB in the irradiation of the 4-CP/TiO2 system also noted that in both cases the major reaction intermediate was 4-CC.17 The results of this work with both TiO2 films and dispersions revealed that the kinetics of photodestruction of 4-CP give a good fit to a LH rate form, over a wide range of different light intensities. These results allowed the extraction of values for kLH and Kapp ads for a TiO2 film and dispersion for different incident light intensities, the results of which are given in Table 1 and illustrated in Figures1 and 2, respectively. An initial analysis2 of these data revealed that kLH and Kapp ads are directly proportional to (a) I1.0 and I-0.5, respectively, when a TiO2 film is used as the photosensitizer, and (b) I0.64 and I-0.32, respectively, when a TiO2 dispersion is used instead. At the time of their production these results were novel, but also hard to explain, without invoking rather involved and possibly unlikely mechanisms.2 Even with the advent of other studies, supporting
14388 J. Phys. Chem. B, Vol. 110, No. 29, 2006
Figure 1. Variation in (a) kLH and (b) Kapp abs , calculated from rate data using eq 2, as a function of a relative incident light intensity, Irel (Irel ) 1 ≡ 6.4 × 1015 photons cm-2 s-1) for the initial photoassisted removal of 4-CP sensitized by a TiO2 film (0.102 mg cm-2, 120.5 cm2). In both a and b the solid lines have been calculated by using eqs 5, 7, and 8 of the pseudo-steady-state model and the optimized fit values, for R, k1, k-1, and θ, given in Table 2.
Figure 2. Variation in (a) kLH and (b) Kapp abs as a function of Irel for the photoassisted removal of 4-CP sensitized by a TiO2 dispersion (0.5 g dm-3; 100 cm3). In both a and b the solid lines have been calculated by using eqs 5, 7, and 8 and the optimized fit values, for R, k1, k-1, and θ, given in Table 2. 10 the above novel discovery that Kapp ads was intensity dependent, 10a and the emergence of possible alternative mechanisms, these results resisted ready interpretation. For example, the revised
Mills et al. LH or Eley-Ridal (ER) based kinetic models of Emeline et al.10a provided an excellent rationale for the dependence they observed -1 of kLH and Kapp ads on I and I , respectively, using phenol as the reactant and P25 dispersions as the photosensitizer. Encouragingly, the variation of Kapp ads and kLH with light intensity for the photoassisted removal of 4-CP observed by Mills and Wang,2 and reported in Table 1, can be fitted very well to the intensity LH model reported by Emeline et al.10a However, a key feature of both the revised LH and ER models is the prediction that 1/ Kapp ads vs kLH should be a straight line with zero intercept and, as we shall see, this does not fit well with the 4-CP data given in Table 1 and illustrated in Figures 1 and 2. In a recent paper,1 Ollis showed that the pseudo-steady-state model gives a good fit to the data reported by Emeline et al. for the photoassisted removal of phenol10a by Degussa P25. It now appears that this same model also provides a reasonable fit to the more challenging data reported by Mills and Wang,2 on the photoassisted removal of 4-CP sensitized by Degussa P25, summarized in Table 1 and illustrated in Figures 1 and 2. Thus, as predicted by eq 7 for TiO2 films, a plot of the values app for Kdiss ()1/Kabsapp) in Table 1 vs Iβ, where β ) 1.0, yields a good straight line from which values for k1/k-1 and R/k1 can be obtained, based on the pseudo-steady-state model. A similar plot, using the results associated with the TiO2 dispersions in Table 1, with β ) 0.6, also generates a good straight line. The values of k1/k-1 and R/k1 derived from these plots are collected in Table 2 and were used, in association with eqs 5, 7, and 8, to generate the solid line model fits to the data shown in Figures 1 and 2. One of the predictions of the pseudo-steady-state model,1 in contrast to other models,10a is that a plot of the data in the form 1/Kabsapp vs k should be a straight line of gradient ) 1/k1 and intercept ) k-1/k1 ) Kdiss ()1/Kads). The data for the TiO2 films and dispersions, given in Table 1 for the photooxidation of 4-CP, were plotted in this form and the results are illustrated in Figure 3, parts a and b, respectively. From these results it would appear that the model, once again, provides a reasonable fit to the data. From the gradient and intercepts of the lines of best fit, values for Kdiss, Kads, k1, and k-1 were obtained for both the TiO2 films and dispersions and are collected in Table 2. From the estimated values of Kads given in Table 2 (6.3 × 103 (films) and 10.8 × 103 M-1 (powder dispersions)) it is clear that they are much greater than that measured for the dark adsorption of 4-CP on TiO2 powder dispersions (1.3 × 102 M-1).14 This is not unusual, as noted earlier by Ollis,1 in reference to his examples,10a,b in his original paper and is due to the reaction sites associated with the process of photosensitization being different, e.g., a small subset of sites or only produced under illumination, to those probed by dark adsorption measurements. From the values of k1 and k-1 in Table 2, it appears that 4-CP is adsorbed more strongly on P25 TiO2 when it is in dispersed, rather than film form, due to a higher rate constant for adsorption, k1. These results suggest that the particle-glass wall interactions and increased particle-particle interactions associated with the films somehow adversely affect the kinetics of adsorption, rather than desorption. The greatest difference between the TiO2 films and dispersions, however, is the dependence upon the incident light intensity, with β equal to 1.0 for films and 0.6 for dispersions. This difference may arise due to a higher threshold at which β changes in value from 1.0 to 0.5 for films, where the particles are permanently intimately linked, compared to dispersions, where they are more isolated. Interestingly, Ahmed et al., using much higher intensities of (2.0-0.7) × 1017 photons cm-2 s-1, have also found that the
Supporting Observations for a Pseudo-Steady-State Model
J. Phys. Chem. B, Vol. 110, No. 29, 2006 14389
TABLE 2: Optimized Fit Parameters to the 4-CP Kinetic Data, Based on the Pseudo-Steady-State Model TiO2 film TiO2 dispersion a
βa
R (µM min-1)b
k-1/k1 (µM)c
k1 (min-1)c
k-1 (µM min-1)c
Kads (103 M-1)c
1.02((0.06) 0.60((0.03)
26.4((0.7) 13.8((0.6)
160(( 30) 92(( 22)
0.054((0.006) 0.087((0.019)
8.6(( 1.9) 8.0(( 2.6)
6.3(( 1.2) 10.8(( 2.6)
Derived from plots of logkLH vs Irel. b Derived from plots of kLH vs Irelθ. c Derived from plots of 1/ Kapp ads vs kLH.
the same authors21 that in most photocatalytic work the free photogenerated carrier concentration in TiO2 will scale almost linearly at absorbed light intensities
